Insert the particles for sedimentation

Submitted by elham.nasiri on Sun, 11/30/2014 - 08:27

Hello everybody
I tried to insert the particles for the creation of the flat bed with the commands as create_atoms or insert/pack but did not succeed!
The created particles do not neighbour enough and the number of those do'nt enough.

Please help me.
Thanks
Elham

Forums:

richti83's picture

richti83 | Sun, 11/30/2014 - 13:03

You can not generate a dense packing with insert/pack, you will only achieve 30% mass fraction because the particles needs to be generated without overlap to respect the history force and torque effects in the contact model.
You have 2 options:
a) generate as much particle as you need using insert/stream and let them settle down
b) use insert pack, but insert with a smaller diameter and than slowly blow the particles up using fix adapt see examples\LIGGGHTS\Tutorials_public\packing\in.packing for an example.

After that you can write a restart file and start your main simulation with a perfect particle bed.

Best,
Christian.

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

mohsen | Tue, 12/02/2014 - 02:37

Hi,

I am running the ErgunTestMPI and I have a problem with particle diameter.
I want to insert two types of material with different density and radius.
I tried to modify in.liggghts_int and in.liggghts_resume to insert my particles.
The problem is that I always get the same particle size/diameter and I don't know about the density.
Following your suggestion I have changed the file according to in.packing as below:

fix pts1 all particletemplate/sphere 1 atom_type 1 density constant 2500 radius constant 0.005
fix pts2 all particletemplate/sphere 1 atom_type 1 density constant 6500 radius constant 0.008
fix pdd1 all particledistribution/discrete 1. 2 pts1 0.3 pts2 0.7
fix ins all insert/pack seed 100001 distributiontemplate pdd1 pdd2 vel constant 0. 0. -1. insert_every once overlapcheck yes all_in yes particles_in_region 10000 region bc

Could you please help me to insert two types of particles.
Best,
Mohsen

richti83's picture

richti83 | Tue, 12/02/2014 - 08:20


fix ins all insert/pack seed 100001 distributiontemplate pdd1 pdd2 vel constant 0. 0. -1. insert_every once overlapcheck yes all_in yes particles_in_region 10000 region bc

should be

fix ins all insert/pack seed 100001 distributiontemplate pdd1 vel constant 0. 0. -1. insert_every once overlapcheck yes all_in yes particles_in_region 10000 region bc

Try to change the order of your ptsX so that the largest particles are inserted as first (because otherwise liggghts inserts small particles first and somtimes there is not enough space for the bigger ones !)

You can dump the mass of each particle when you add keyword mass to your dump command, than you can calculate the density.

Best,
Christian

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

mohsen | Wed, 12/03/2014 - 06:14

Thanks for your prompt reply. I tried the following syntax. But again I got particles with 0.0005 m diameter and 2000 density which is not listed below.

#particle distributions and insertion
region bc cylinder z 0.0 0.0 0.012 0. 0.055 units box
fix pts1 all particletemplate/sphere 1 atom_type 1 density constant 6500 radius constant 0.005
fix pts2 all particletemplate/sphere 1 atom_type 1 density constant 2500 radius constant 0.004
fix pdd1 all particledistribution/discrete 1. 2 pts1 0.3 pts2 0.7
fix ins all insert/pack seed 100001 distributiontemplate pdd1 vel constant 0. 0. -1. insert_every once overlapcheck yes all_in yes particles_in_region 10000 region bc

I inserted the density in in.liggghts_resume to keep an eye on the density of the particles and it worked.

elham.nasiri | Sun, 11/30/2014 - 13:16

Hi Christian
Thanks for your reply. I will try these methods.

With Best Wishes
Elham

elham.nasiri | Tue, 12/02/2014 - 09:15

Hi Christian
I used insert/stream for insertation of particles but after settleing, the surface of particles did'nt uniform for flat bed and they did'nt dense!

Thanks
Elham

richti83's picture

richti83 | Tue, 12/02/2014 - 09:37

What is your coefficent of restitution, friction, rolling friction, youngmoduls and dt ?

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

elham.nasiri | Wed, 12/03/2014 - 06:56

Hi Christian
Thanks. you're right. The pattern of setteling changed with changing the coefficients.

Best
Elham

elham.nasiri | Sun, 12/07/2014 - 08:00

Hi christian
I have the problems yet...
I used below timestep and coefficients but the particles did not be compact and dense, and surface of bed did not flat. The diameter and density of quartz particles is 0.7 mm and 2650 Kg/m3.
{
timestep 0.000001

variable yMp1 equal 700e9
variable yMwall equal 90e9

variable pRp1 equal 0.18
variable pRwall equal 0.3

variable cRp1p1 equal 0.75
variable cRp1wall equal 0.5
variable cRwallp1 equal 0.5
variable cRwallwall equal 0.7

variable cFp1p1 equal 0.65
variable cFp1wall equal 0.45
variable cFwallp1 equal 0.45
variable cFwallwall equal 0.7
}

what is the effect of any coefficient in setteling?

Also I have other guestion, what is the reason of below error?
WARNING: Fix insert/stream: Extruded insertion face extends outside domain, may not insert all particles correctly (../fix_insert_stream.cpp:468)
ERROR: Fix insert/stream (id ins): insertion volume could not be distributed properly in parallel. Bad decomposition or insertion face extrusion is too small or outside domain (../fix_insert_stream.cpp:459)

Thanks
Elham

elham.nasiri | Tue, 12/09/2014 - 13:30

Hi again
I guess it's not possible to get dense particles with real porosity. Is it true?

Best

richti83's picture

richti83 | Tue, 12/09/2014 - 18:05

Hi,

I'm not sure how deep your DEM background is,but I can tell you that you can not use the real YoungModulus (700e9 in your example) because the critical timestep decreases significantly. (t_krit=sqrt(m/k)).
We usually set Y=1e7 .. 5e7, this leads to very soft behaivour and high overlap while the contact forces are similar to very stiff particles (with less overlap) because F=k*x.

On the oder hand, I think the classic hertz model is not a good model for a dense packing, because of the 'restoring force'. I can tell you that I never reached porosity lower than 0.4. There are other models (hysteresis, Luding, Edinburgh-Elasto-Plastic Adhesion) which uses a hysteresis in relaxing part of contact model so that the particles keep overlaped when the outer force is removed. (see this thread about some more information and links: http://www.cfdem.com/forums/about-normal-model-hooke-hysteresis) I would like to write a paper about this topic next year to give some deeper details, but first need to write some other things ...

I cant not really decode your variables, but when 0.75 is your coefficent of rolling friction than it is very high, and your particles will behave like a 'qube'. See Wensrich & Katterfeld "Rolling friction as a technique for modelling particle shape in DEM"
I would try it with Y=1e7, rollingFricParticleParticle = 0.2, Friction = 0.3, CoefficentOfRestition=0.2.

About your warning:
WARNING: Fix insert/stream: Extruded insertion face extends outside domain, may not insert all particles correctly (../fix_insert_stream.cpp:468)
Your simulation region is smaller than the volume above your insert mesh (decrease extrusion length or increase box-height)
Lets say your fix mesh/surface for insert stream is in z=1, your extrusion length is 0.3, than your region reg for create_box command need to have xLo xHi yLo yHi zLOw 1.3 #[m]

Best,
Christian

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

elham.nasiri | Wed, 12/10/2014 - 09:10

Hi,

Thanks for your comprehensive reply.
I don't have high information in regard to DEM. I studied some papers and used these coeffecients. My rolling friction was 0.1 not 0.75 that forgeted in writing the coffecients for you.
Totally, I'm confused about the simulation of scouring. Is it neccesary the conditions of bed particles and their density be same with real sediment or not?

Aside these, the surface of particles after settleing was not flat! why? is it necessary to smooth it?

About my warning, I changed size of simulation box but there was same warning!

Best,
Elham

mohsen | Wed, 01/07/2015 - 01:30

Hi Elham,

I am using the ErgunTest for sedimentation. I had problem with Young modulus just like your case. I change the time step for DEM to 0.000001 and the young modulus is now 1e9 my simulation is working well. I am not sure if it is a good solution to the problem.
If you are using Paraview to visualize your simulation and particles are not smooth maybe you should change the Theta Resolution to a higher value in Glyph settings. This just increases the quality of your visulization and has nothing to do with the answer.
I hope this can solve your problem. I would be happy if we can share our experince about this.
I still have problem with max packing. I think I can not achieve the required packing.

Best,
Mohsen

elham.nasiri | Sun, 05/24/2015 - 12:42

Hi Christian,
Good times,

I'm going to fill a subspace with particles and then replicating this through the full volume. I inserted the particles by insert/stream command and let them to settle with "fix nve" command. But after useing replicate command for duplicating them, there is a error, can't replicate because of fix atoms properties.

Please let me know if you know why!
Thanks
Elham